http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
심종성 ( Sim Jongsung ),문도영 ( Moon Doyoung ),오홍섭 ( Oh Hongseob ),하우진 ( Ha Woojin ) 한국구조물진단유지관리공학회 2004 한국구조물진단유지관리공학회 학술발표대회 논문집 Vol.8 No.2
In this study, pull-out test of GFRP reinforcing bars with deformed type rib, which is contained with milled fiber were performed. Deformed type rib is attached to surface of FRP re-bar for improving bond performance to concrete and formability of surface. The maximum average bond strength, stiffness and failure mode of pull-out test specimens reinforced with GFRP re-bar were evaluated. As a results of test, it is confirmed that short fiber in rib increased maximum bond strength. Also, basic development length of GFRP re-bar in accordance with ACI 440. 1R-26 is calculated and compared with experimental results.
심종성(Sim Jongsung),문도영(Moon Doyoung),오홍섭(Oh Hongseob),하우진(Ha Woojin) 한국구조물진단유지관리학회 2004 한국구조물진단학회 학술발표회논문집 Vol.- No.-
In this study, pull-out test of GFRP reinforcing bars with deformed type rib, which is contained with milled fiber were performed. Deformed type rib is attached to surface of FRP fe-bar for Improving bond performance to concrete and formability of surface. The maximum average bond strength, stiffness and failure mode of pull-out test specimens reinforced with GFRP fe-bar were evaluated As a results of test, It is confirmed that short fiber in rib increased maximum bond strength Also, basic development length of GFRP re-bar in accordance With ACI 440 lR-26 18 calculated and compared with experimental results.
심층 신경망 기반 효율적인 단일 영상 초해상도 복원 기법
정우진(Woojin Jeong),양현석(Hyeon Seok Yang),한복규(Bok Gyu Han),심재준(Jae Jun Sim),박세진(Sejin Park),박진욱(Jin Wook Park),이종민(Jong Min Lee),문영식(Young Shik Moon) 대한전자공학회 2018 전자공학회논문지 Vol.55 No.6
단일 영상 초해상도 복원은 하나의 저해상도 영상에서 고해상도 영상을 복원하는 과정이다. 최근 깊은 인공 신경망 기술이 발전함에 따라 단일 영상 초해상도 복원에서도 깊은 인공 신경망 기술이 성과를 나타냈다. 본 논문은 단일 영상 초해상도 복원을 위해 깊은 인공 신경망 기술을 효율적으로 적용하는 방법에 대해 연구하였으며, 네트워크 내부 확대 기법, L1 손실 함수의 사용, 잔차 학습 구조를 통해 기존 기법보다 효율적으로 영상 복원하는 기법을 제안한다. 제안하는 방법은 기존 방법보다 화질은 PSNR기준으로 0.57㏈ 만큼 우수하며 속도는 1.48배 빠른 것을 실험을 통해 확인하였다. Single image super-resolution is to restore a high-resolution image from a low-resolution image. Recently, deep neural networks have been applied in various image processing field, and they achieve successful results in the single image super-resolution. In this paper, we propose an efficient way of utilizing the deep neural networks to the single image super-resolution. we improve the quality of single image super-resolution by using convolution transpose layer, L1 loss function, and residual learning. Experimental results have shown that our method is 0.57㏈ better in terms of PSNR and 1.48 times faster in execution time, compared with existing methods.